Scan 1

Gray chiton, Lepidochitona cinerea (Polyplacophora: Lepidochitonidae), adult, 15.5 mm total length (TL), formalin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in ethanol (Table 2). A digitally reconstructed radiography (DRR) rendering illustrates that the chosen scan settings permit recognition of hard parts only, despite the presence of soft tissues in this ethanol-preserved sample (Fig. 1A). Virtual sections and volume renderings at different positions and from various viewpoints show that mineralized tissues such as valves (Fig. 1B), spiculae (Fig. 1C), radula (Fig. 1D), tegmentum (Fig. 1E), and pustules (Fig. 1F), but also X-ray-dense ingesta inside the posterior digestive tract (Fig. 1G) can be visualized using µCT. A close-up view of the radula (Fig. 1H) reveals that even individual radular teeth can be successfully differentiated at the given resolution. Because of its relatively large size, the dataset was binned (i.e., down-sampled) from its original 8 µm to 16 µm isotropic voxel resolution (IVR) for online deposition. Due to the 3D nature of the tomographic data, this reduction in resolution by a factor of two results in a reduction of the dataset's volume by a factor of eight (i.e., 2³). While binning lowers resolution, it may result in an improved signal-to-noise ratio (SNR), thus artificially improving the visual aspect of the dataset.

Figure 1.

Micro-computed tomography data of an adult specimen of the polyplacophoran Lepidochitona cinerea. A. Digitally reconstructed radiograph rendering, lateral view with anterior to right. B. Virtual sagittal section at the level of the radula. C. Solid volume rendering. D. Virtual sagittal slicing of a solid volume rendering at the level of the radula. E. Virtual transverse section, position indicated by white arrow as in C. F. Solid volume rendering, anterior view. G. Solid volume rendering, ventral view with anterior to right. H. Close-up view of the radula. All renderings and virtual sections created using the commercial software Amira (A) and VGSTUDIO MAX (B-H) at 8 µm isotropic voxel resolution.

Scan 2

Northern chiton, Leptochiton asellus (Polyplacophora: Leptochitonidae), adult, 13 mm TL, Bouin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in air (Table 2). For high-resolution scanning of an isolated chiton body part, the anterior valve was removed and air-dried. This scan was made to obtain an overview of the entire sample prior to scanning at higher resolutions (see below). A DRR rendering partly reveals the internal macrostructure of the anteriorvalve (Fig. 2A), while a solid volume rendering can be used to show the superficial distribution of the sensory pores (Fig. 2B). However, a semi-transparent rendering (Fig. 2C) as well as a consecutive series of a virtual sagittal section (Fig. 2D), solid (Fig. 2E), and sliced (Fig. 2F, G) volume renderings are necessary to obtain a more comprehensive understanding of the underlying megalaesthete morphology. Despite the insight provided by this scan, a higher IVR is required to reveal smaller structures, specifically the micraesthetes.

Figure 2.

Micro-computed tomography data of the anterior valve of an adult specimen of the polyplacophoran Leptochiton asellus. The two datasets were acquired at 8 µm (A-G) and 600 nm (H-J) isotropic voxel resolution. A. Digitally reconstructed radiograph rendering, dorsal view with anterior facing up. B. Solid volume rendering. C. Semi-transparent volume rendering. D. Virtual sagittal section, anterior to left, position indicated by white arrow as in C. E. Solid volume rendering, lateral view. F. Virtual sagittal slicing of a solid volume rendering. G. Close-up as indicated in F. H. Close-up of the approximate area as indicated in D. I. Semi-transparent volume rendering of the distal area of a single megalaesthete. J. Dorsal view of part of the anterior valve. All renderings and virtual sections created using the commercial software Amira (A) as well as the free software Drishti (B, C, E-G, I, J) and Fiji (D, H).

Scan 3

Northern chiton, Leptochiton asellus (Polyplacophora: Leptochitonidae), adult, 13 mm TL, Bouin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in air (Table 2). In order to obtain a dataset with sub-micron resolution, only the anterior tip of the valve was scanned. As a virtual sagittal section through the resulting tomographic data shows, micraesthetes and megalaesthetes can now be differentiated (Fig. 2H). Close-up sagittal (Fig. 2I) and dorsal views (Fig. 2J) of the scanned region of interest (ROI) provide further insight into sensory pore morphology of this polyplacophoran species.

Scan 4

Common cockle, Cerastoderma edule (Bivalvia: Cardiidae), adult, 21 mm TL, Bouin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in ethanol (Table 2). As in the first scan (Fig. 1), the scanning parameters chosen here do not permit any differentiation of soft parts, despite their presence in this ethanol-preserved sample. This lack of soft tissue contrast is illustrated by a DRR rendering (Fig. 3A) as well as a virtual transverse section (Fig. 3B). However, despite its significantly lower density compared to the two valves, the ligament can still be identified (Fig. 3C). In addition, the previous as well as a further volume rendering permit obtaining an unobstructed view of all major hard parts, including the articulated hinge as seen from an internal viewpoint (Fig. 3D). These latter two images illustrate that visualization software can be used to generate false-colored renderings that closely approach the aspect of the specimen in real life (Fig. 3C).

Figure 3.

Micro-computed tomography data of adult specimens of the bivalve Cerastoderma edule (A-D), the gastropod Cerithium litteratum (E-H), and the cephalopod Nautilus pompilius (I-L). A, E, I. Digitally reconstructed radiograph rendering, lateral view. B, F, J. Virtual transverse section. C, G, K. Solid volume rendering. D, H, L. Virtual slicing of a solid volume rendering, oblique view. All renderings and virtual sections created using the commercial software Amira (A, E, I, K, L) and VGSTUDIO MAX (G, H) as well as the free software Fiji (B, F, J) and Drishti (C, D) at 26 µm (A-D), 25 µm (E-H), and 84.7 µm (I-L) isotropic voxel resolution.

Scan 5

Stocky cerith, Cerithium litteratum (Gastropoda: Cerithiidae), adult, 22.5 mm TL, dry shell, dataset acquired ex vivo, scanned in air (Table 2). A DRR rendering shows that the intricate internal coiling of the snail's shell can be easily visualized using µCT (Fig. 3E). However, a virtual horizontal section reveals that the shell’s microstructure cannot be differentiated at the given resolution (Fig. 3F). Two volume renderings illustrate the capacity of modern visualization tools to provide a sense of depth through virtual shading (Fig. 3G) and slicing (Fig. 3H).

Scan 6

Chambered nautilus, Nautilus pompilius (Cephalopoda: Nautilidae), adult, 77 mm TL, dry shell, dataset acquired ex vivo, scanned in air (Table 2). A DRR rendering provides insight into the complex internal morphology of the chambered nautilus' shell (Fig. 3I). In addition, virtual sectioning can be used to generate views typically employed in nautilid taxonomy and systematics without the need to alter the specimen (Fig. 3J). False-colored volume renderings further illustrate that conventional (Fig. 3K) as well as entirely novel perspectives (Fig. 3L) of the nautilid shell can be obtained using digital µCT data.

Scan 7

Humboldt squid, Dosidicus gigas (Cephalopoda: Ommastrephidae), adult, 1.55 m TL, frozen specimen, thawed prior to scanning, dataset acquired ex vivo, scanned in air (Table 2). A DRR rendering from a dorsal viewpoint illustrates the overall shape and scanning position of the animal (Fig. 4A). A virtual section at the level of the large, centrally located digestive gland reveals selected soft tissue organs such as gills or optic lobes, but also hard parts such as the statoliths (Fig. 4B). An impression of the external morphology can be obtained by solid volume rendering of the entire dataset (Fig. 4C), while selected internal organs can best be visualized using a semitransparent rendering (Fig. 4D). Note the strongly X-ray absorbing content of the ink sac in contrast to the relatively weak attenuation of gladius as well as lower and upper beak (Fig. 4D).

Figure 4.

Computed tomography data of an adult specimen of the cephalopod Dosidicus gigas. A. Digitally reconstructed radiograph rendering, dorsal view with anterior facing right. B.Virtual horizontal section at the level of digestive gland and gills. C. Solid volume rendering showing external anatomical features. D. Semi-transparent volume rendering exposing selected internal organs. All renderings and virtual sections produced using the commercial software Amira at 683 µm isotropic voxel resolution.

Scan 8

Spiny chiton, Acanthochitona crinita (Polyplacophora: Acanthochitonidae), adult, 32 mm TL, Bouin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in ethanol (Table 2). Lateral and ventral views based on volume renderings of the 3D dataset show that all external hard and soft part features of this chiton such as valves, girdle, foot, or gills can be readily visualized at the given resolution (Fig. 5A, B). In addition, a virtual transverse section immediately posterior to the head demonstrates that major internal organ systems can be differentiated as well (Fig. 5C). In this particular image, individual lacunae located in the girdle and foot as well as the complex muscular disposition of the radula apparatus become apparent.

Scan 9

Ram's horn squid, Spirula spirula (Cephalopoda: Spirulidae), adult, 63 mm TL, formalin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in ethanol (Table 2). A DRR rendering from a lateral viewpoint suggests that the contrast agent has permeated the entire specimen, thus rendering all internal structures visible (Fig. 5D). Consequently, a virtual sagittal section reveals not only the hard part morphology of the animal with its coiled chambered shell, but also its soft part composition with buccal mass, statocysts, central nervous system, and various other internal organs (Fig. 5E). In addition, the 3D dataset can be used to visualize the external aspect of the animal with arms, tentacles, eyes, mantle, and fins (Fig. 5F). However, as a virtual slicing of a volume rendering seen from a lateral viewpoint shows, a small area at the posterior end of the digestive gland has not been stained (Fig. 5G). Despite several weeks of immersion in the contrast agent and occasional flushing of the mantle cavity, diffusion of PTA into this part of the sample was not achieved.

Figure 5.

Micro-computed tomography data of contrast-enhanced adult specimens of the polyplacophoran Acanthochitona crinita (A-C) and the cephalopod Spirula spirula (D-G). A. Solid volume rendering, lateral view with anterior facing right. B. Solid volume rendering, ventral view. C. Virtual transverse section, position indicated by white arrow in (A). D. Digitally reconstructed radiograph rendering, lateral view with anterior to right. E. Virtual sagittal section at the level of the right statocyst. F. Solid volume rendering. G. Virtual slicing of a solid volume rendering at the level of the digestive gland. Asterisk indicates region with insufficient penetration of the contrast agent. All renderings and virtual sections created using the free software Drishti (A, B) and Fiji (C) as well as the commercial software Amira (D-G) at 24 µm (A-C) and 40 µm (D-G) isotropic voxel resolution.

Scan 10

Inkless octopus, Muusoctopus sp. (Cephalopoda: Enteroctopodidae), embryo, 12 mm TL, formalin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in ethanol (Table 2). This specimen was part of an egg clutch collected during a scientific cruise in the Northern Pacific Ocean (Werner et al. 2016). The embryo was removed from its egg, because the contrast agent was incapable of penetrating the egg envelope. A DRR rendering shows that in addition to the external morphology, several internal structures of the specimen are visible, including the eyes and digestive gland (Fig. 6A). A virtual horizontal section reveals that major organ systems such as digestive tract, sensory organs, nervous system, or respiratory structures can be differentiated without difficulty (Fig. 6B). Virtual slicing of the volume-rendered dataset in combination with a 2D overlay permit obtaining a more plastic impression of the animal's internal structural composition (Fig. 6C). Furthermore, volume renderings of the dataset allow visualizing various external structural features such as the arms, suckers, or mantle (Fig. 6D). Due to the 3D nature of the dataset, virtual slicing can be used to create any desired arbitrary section, thus revealing additional internal organs such as the systemic heart, the statocysts, or individual components of the central nervous system (Fig. 6E). In addition, selected internal organs were manually segmented and surface-rendered, an approach that permits precise volumetric analyses and visualization of selected structures (Fig. 6F). Note the relatively high number of averages (Table 2), a parameter that was chosen to improve SNR in this scan of a contrasted animal scanned in ethanol compared with non-contrasted specimens scanned in air (e.g., scan 2). Due to the X-ray attenuation of the liquid surrounding the specimen, multiple averages or longer exposure times need to be applied to obtain a better SNR.

Figure 6.

Micro-computed tomography data of a contrast-enhanced embryo of the cephalopod Muusoctopus sp. A. Digitally reconstructed radiograph rendering, dorsal view with anterior facing right. B. Virtual horizontal section at the level of the eyes. C. Virtual horizontal slicing of a solid volume rendering. D. Solid volume rendering, lateral view with anterior facing right. E. Virtual sagittal section at the level of the systemic heart. F. Digitally reconstructed radiograph rendering illustrating position of segmented internal organs, oblique anterodorsal view. Branchial gland = dark green, branchial heart = light brown, digestive tract = blue, eye = orange, gill = light green, nervous system = yellow, posterior salivary gland = light blue, white body = white. All renderings and virtual sections created using the commercial software Amira at 10 µm isotropic voxel resolution.

Scan 11

Common cockle, Cerastoderma edule (Bivalvia: Cardiidae), adult, 21 mm TL, formalin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in distilled water (Table 3). Virtual sections illustrate that a number of internal organs can be discerned using preclinical MRI, including the foot, adductor and retractor muscles, ligament, and various elements of the digestive tract (Fig. 7A, B). These structures can additionally be volume-rendered due to the relatively weak signal intensity of the surrounding water, a consequence of the specific MRI protocol chosen here (Fig. 7C). Note the susceptibility artifacts at the intersection of valves and the surrounding liquid - such distortions often occur at boundaries between tissues with different chemical properties (Fig. 7B).

Scan 12

Blue mussel, Mytilus edulis (Bivalvia: Mytilidae), adult, 71 mm TL, dataset acquired in vivo, scanned in seawater (Table 3). The imaging protocol employed here was a rapid 2D sequence with an acquisition time of slightly more than a minute. A series of orthogonal sections illustrates that even with such a reduced scanning time and relatively low resolution, all major internal organs of a common bivalve can be differentiated, in particular gills, adductor and retractor muscles (Fig. 7D) as well as mantle (Fig. 7E) and various components of the digestive tract (Fig. 7F).

Scan 13

Bushy-backed nudibranch, Dendronotus frondosus (Gastropoda: Dendronotidae), adult, 24.5 mm TL, formalin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in distilled water (Table 3). A virtual sagittal section through the dataset along the median plane shows that various internal and external structures can be differentiated, including the digestive tract, gonad, and foot (Fig. 7G). In order to test whether this scan could also be used for volumetric analyses, the original 2D dataset was zero-filled (i.e., increased in resolution) from 313 µm to 81 µm along the Z axis, which resulted in a 3D dataset with 81 µm IVR. Visualization of this new image stack in the form of a virtually sliced volume rendering shows that the artificial increase in resolution along the third dimension can be used as a tool to acquire volumetric information from 2D MRI data (Fig. 7H). However, this artificial increase in resolution along the Z axis obviously cannot provide any additional structural information.

Figure 7.

Magnetic resonance imaging data of adult specimens of the bivalves Cerastoderma edule (A-C) and Mytilus edulis (D-F) as well as the gastropods Dendronotus frondosus (G, H) and Carronella pellucida (I) obtained using high-field preclinical scanning systems. A. Virtual sagittal section at the level of the intestine, anterior facing left. B. Virtual transverse section at the level of the kidney. Asterisks denote susceptibility artifacts. C. Semi-transparent volume rendering, lateral view. D. Virtual sagittal section at the level of the stomach, anterior facing right. E. Virtual transverse section at the level of the foot. F. Virtual horizontal section at the level of the adductor, anterior facing right. G. Virtual sagittal section at the level of the mouth, anterior facing left. H. Virtual sagittal slicing of a solid volume rendering, oblique anterodorsal view. I. Virtual sagittal section at the level of the digestive gland, anterior to left. All renderings and virtual sections created using the free software Fiji (A-C), the proprietary scanner software ParaVision (D-F, I), and the commercial software Amira (G, H) at 40.5 µm isotropic (A-C), 39.3 × 39.3 × 2,000 µm (D-F), 40.5 µm isotropic (G, H), and 164 × 164 × 500 µm (I) voxel resolution.

Scan 14

Red-gilled nudibranch, Carronella pellucida (Gastropoda: Flabellinidae), adult, 25 mm TL, Bouin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in distilled water (Table 3). A single virtual sagittal section illustrates that in less than a quarter of an hour all major internal structures as well as external structures of a nudibranch can be visualized using preclinical MRI (Fig. 7I).

Scan 15

Scorched mussel, Brachidontes exustus (Bivalvia: Mytilidae), juvenile, 16 mm TL, ethanol-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in agarose (Table 3). A virtual sagittal section shows that the chosen scanning parameters permit unambiguous identification of a considerable amount of internal structures such as the adductor muscles, mesosoma, stomach, and pericard (Fig. 8A). Further internal organs that can be observed are elements of the digestive tract, respiratory structures, and circulatory system (Fig. 8B), but also components of the excretory and reproductive systems (Fig. 8C). In this particular scan, a surface instead of a volume coil was used, which resulted in a notable shading gradient within the dataset. The source of this particular artifact is the surface coil itself, which has an optimal sensitivity at the center of the field of view (FOV) and therefore can cover only a certain part of the measured volume with homogenous signal intensity. However, such shading gradients can usually be removed from 2D as well as 3D tomographic image stacks by employing the Polynomial Shading Corrector, a plugin available for Fiji.

Figure 8.

Magnetic resonance imaging data of adult specimens of the bivalves Brachidontes exustus (A-C), Ostrea edulis (D, E), Crassostrea virginica (F), and Pecten maximus (G) as well as the cephalopod Sepia officinalis (H, I) obtained using high-field preclinical scanning systems. A. Virtual sagittal section, anterior to right. B. Virtual transverse section. C. Virtual horizontal section, anterior to right. D. Virtual sagittal section at the level of the gills, anterior to right. Asterisk indicates a smaller specimen attached to the animal originally studied. E. Virtual horizontal section at the level of the adductor. F. Virtual sagittal section through a region of interest surrounding heart and adductor muscle, anterior to right. G. Virtual sagittal section, anterior facing right. H. Virtual horizontal section at the level of the statocysts, anterior to right. I. Composite figure showing sagittal, transverse, and horizontal false-colored virtual sections, oblique anterodorsal view. All renderings and virtual sections created using the free software Fiji (A-C), the proprietary software ParaVision (D-G), and the commercial software Amira (H, I) at 20 µm isotropic (A-C), 234 × 234 ×; 1,000 µm (D, E), 625 × 625 × 2,740 µm (F), 234 × 234 × 1,000 µm (G), and 33 µm isotropic (H, I) voxel resolution.

Scan 16

European flat oyster, Ostrea edulis (Bivalvia: Ostreidae), adult, 143 mm TL, dataset acquired in vivo, scanned in sea water (Table 3). Similar to the specimen of M. edulis (scan 12), virtual sections through this 2D dataset illustrate that despite the short acquisition times typically used for in vivo MRI, all major internal organ systems can be identified with a high level of confidence (Fig. 8D, E). Such data are of particular interest when respiratory or muscular performance has to be evaluated rapidly and non-invasively using living specimens. Note in particular the difference in contrast within the adductor muscle, which illustrates its bipartite macrostructure divided into smooth (catch) and striated (quick) muscle (Fig. 8D, E). The scan also reveals that a smaller, juvenile specimen is attached to the initially analyzed adult oyster (Fig. 8D).

Scan 17

Eastern oyster, Crassostrea virginica (Bivalvia: Ostreidae), adult, 135 mm TL, dataset acquired in vivo, scanned in sea water (Table 3). For this particular specimen, a ROI focusing on the viscera was chosen. As a single virtual sagittal section shows, the data permit differentiation of the main muscular and circulatory elements (Fig. 8F).

Scan 18

Great scallop, Pecten maximus (Bivalvia: Pectinidae), adult, 99 mm TL, dataset acquired in vivo, scanned in sea water (Table 3). A virtual sagittal section shows that similar to the scans performed on living specimens of the bivalve species M. edulis (scan 12), O. edulis (scan 16), and C. virginica (scan 17) all major internal organ systems can be recognized (Fig. 8G). Particularly conspicuous are the digestive gland, kidney, gills, and adductor muscle. In addition, the shape of the mineralized valves can be ascertained due to the positive contrast provided by the signal-emitting liquid (i.e., sea water) surrounding these structures. However, these in vivo MRI data show more movement artifacts around the gills than in the oyster datasets due to a generally increased water circulation inside scallops.

Scan 19

Common cuttlefish, Sepia officinalis (Cephalopoda: Sepiidae), sub-adult, 37 mm TL, formalin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in distilled water (Table 3). A virtual horizontal section shows that cephalopod internal anatomy can be visualized at a high level of confidence using preclinical MRI (Fig. 8H). Prominent internal structures that can be identified are the digestive gland, ink sac, optic lobes, buccal mass, and statocysts, while external structures such as suckers or the muscular mantle can be seen as well. In addition, smaller anatomical structures like the lacunae or individual components of the statocysts can be differentiated as well. These structures can also be seen in arbitrary 2D sections through the 3D dataset, here shown in the form of a false-colored composite image of a horizontal, sagittal, and transverse section (Fig. 8I).

Scan 20

Ram's horn squid, Spirula spirula (Cephalopoda: Spirulidae), adult, 55 mm TL, formalin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in distilled water (Table 3). As virtual horizontal and transverse sections through the MRI dataset illustrate, various internal organs can be identified in this specimen. For example, due to the divergent contrast properties of the constituent as well as the surrounding tissues, arm nerves or the esophagus can be readily singled out (Fig. 9A, B). For segmentation and subsequent 3D modeling purposes, the 2D dataset was zero-filled to 60 µm along the Z axis. As an oblique anterodorsal view of the resulting 3D model shows, segmentation of complex MRI datasets can be used to single out individual structures for their more informative presentation in a didactical sense (Fig. 9C).

Scan 21

Common cuttlefish, Sepia officinalis (Cephalopoda: Sepiidae), sub-adult, 37 mm TL, formalin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in distilled water (Table 3). This 2D scan was made to obtain a morphological reference dataset for the subsequently conducted DTI scan (see below). As was the case for all previous MRI scans of cephalopods (Figs. 8H, 9A), the acquired data permit rapid differentiation of major internal and external structures (Fig. 9D).

Scan 22

Common cuttlefish, Sepia officinalis (Cephalopoda: Sepiidae), sub-adult, 37 mm TL, formalin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in distilled water (Table 3). As a proof-of-principle for the application of DTI in malacological research, brain and optic lobes were selected for fiber tract visualization in a cuttlefish. To this end, a ROI comprising the head was chosen (Fig. 9E). The data show that diffusion along fiber tracts is still present post mortem (Fig. 9F) and that density of the fiber tracts is significantly higher in the optic lobes than in the brain (Fig. 9G).

Figure 9.

Magnetic resonance imaging data of adult specimens of the cephalopods Spirula spirula (A-C) and Sepia officinalis (D-G) obtained using high-field preclinical scanning systems. A. Virtual horizontal section, anterior to right. B. Virtual transverse section. C. Solid and semi-transparent surface rendering of segmented structures, oblique anterodorsal view. Blue = digestive tract, brown = digestive gland, green = statocysts, grey = mantle, red = eyes, yellow = nervous system. D. Virtual horizontal section at the level of the optic lobes, anterior to right. Areas selected for visualization by diffusion tensor imaging segmented in red. E. Semi-transparent volume rendering of the areas segmented as in D. shown in combination with a virtual transverse section at the level of the lenses, oblique posterodorsal view. F, G. Diffusion tensor imaging data showing fiber tract orientation in optic lobes and the brain, dorsal view. All renderings and virtual sections created using the free software Fiji (A, B, D), Acrobat Reader (C), and DSI Studio (E-G) at 30 µm isotropic (A-C), 30 × 30 × 100 µm (D), and 58.6 × 58.6 × 235 µm (E-G) voxel resolution.

Scan 23

Oman cuttlefish, Sepia omani (Cephalopoda: Sepiidae), adult, 247 mm TL, formalin-fixed, ethanol-preserved, dataset acquired ex vivo, scanned in air (Table 3). This particular specimen was removed from its ethanol-filled container, rinsed, and placed inside an air-filled glass jar that was sealed prior to positioning inside the MRI coil. Using this experimental setup, the evaporation of the ethanol still contained within the sample's tissues led to an alcohol-saturated atmosphere inside the glass jar, effectively preventing specimen movement as a result of desiccation during the scan. In addition, structural integrity of the animal during the scan was ensured due to the rubber-like properties of its formalin-fixed and long-term ethanol-preserved soft tissues. A virtual horizontal section at the level of the eyes shows that various external and internal structures can be differentiated, including the mantle, gonad, digestive gland, gills, statocysts, lenses, and buccal mass (Fig. 10A). Due to the specimen being surrounded by air, the strong contrast between the object and the surrounding medium results in the possibility to obtain informative volume rendering data of the entire specimen without the need for manual segmentation of individual structures (Fig. 10B).

Figure 10.

Magnetic resonance imaging data of adult specimens of the cephalopods Sepia omani (A, B), Sepia officinalis (C-E), and Grimpoteuthis sp. (F, G) obtained using high-field clinical scanning systems. A. Virtual horizontal section at the level of the statocyst, anterior to right. B. Semi-transparent volume rendering, oblique anterodorsal view. C. Virtual horizontal section at the level of the digestive gland, anterior facing right. D. Virtual horizontal section at the level of the eyes, anterior facing up. E. Semi-transparent volume rendering, dorsal view. F. Virtual horizontal section at the level of the optic lobes, anterior to right. G. Solid volume rendering of selected internal organs combined with a virtual sagittal section, oblique anterodorsal view. All renderings and virtual sections created using the commercial software Amira (A, B, F, G), the free software Fiji (C), and the proprietary scanner software syngo (D, E) at 173.5 µm isotropic (A, B), 200 × 200 × 1,000 µm (C), 117 µm isotropic (D, E), and 140 µm isotropic (F, G) voxel resolution.

Scan 24

Common cuttlefish, Sepia officinalis (Cephalopoda: Sepiidae), adult, 341 mm TL, dataset acquired in vivo, scanned partially immersed in sea water (Table 3). As a virtual horizontal section shows, these rapidly generated 2D MRI data permit differentiation of all major internal organ systems, including the gonads, components of the digestive tract, gills, eyes, and lenses (Fig. 10C). In addition, mineralized and therefore per se not very MRI-signal-intense structures such as the lower and upper beak or lenses become apparent as well due to being surrounded by tissues rich in signal-emitting water molecules. A direct comparison with the data from the fixed specimen of S. omani shown in the previous scan (Fig. 10A) illustrates a considerable degree of distortion that specific cephalopod organs may undergo during fixation and subsequent long-term storage in ethanol.

Scan 25

Common cuttlefish, Sepia officinalis (Cephalopoda: Sepiidae), adult, 341 mm TL, dataset acquired in vivo, scanned partially immersed in sea water (Table 3). The living, adult specimen of S. officinalis introduced in the previous scan was analyzed again during the same imaging session using a ROI focusing on the specimen s head. A virtual horizontal section illustrates that the resulting data permit differentiating white bodies from the optic lobes as well as selected internal elements of the statocysts (Fig. 10D). Due to the 3D nature of the acquired dataset, volume rendering allows obtaining a more plastic impression of the structural composition of the specimen's head (Fig. 10E).

Scan 26

Dumbo octopus, Grimpoteuthis sp. (Cephalopoda: Grimpoteuthidae), adult, 290 mm TL, formalin-fixed, formalin-preserved, dataset acquired ex vivo, scanned in formalin (Table 3). This intact cirrate specimen was collected during a scientific cruise in the Northern Pacific Ocean (Werner et al. 2016). As a virtual horizontal section shows, all internal structures relevant for the identification of a dumbo octopus to species level can be differentiated (Fig. 10F). These structures include, among others, digestive tract, central nervous system with optic lobes and nerves, white bodies, arms, web, fins and their muscles, as well as eyes. All of these organs are suitable for 3D reconstruction using manual segmentation followed by surface rendering. In addition, a volume rendering of selected internal organs reveals further structures that can be visualized in 3D, including the gills, branchial hearts, as well as all major components of the digestive tract such as esophagus, stomach, caecum, and rectum (Fig. 10G). The presence of a large testis shows that the captured animal is an adult male (Fig. 10G). In contrast to the previous two cephalopod specimens analyzed using clinical MRI (scans 23-25), this sample was scanned fully immersed in a liquid rich in MRI-signal-emitting water molecules (i.e., formalin), which complicates threshold-based volume rendering due to the absence of a sharp boundary between specimen and surrounding medium.

µCT

As demonstrated here, X-ray-based techniques are particularly suited for the rapid visualization of molluscan hard part morphology in 3D, including spicules, radula, and ligament, but in particular aspects of the macro- and microstructure of the molluscan shell (Figs. 1-3). The possibility to generate such data using µCT had already been noticed more than a decade ago (Vendrasco et al. 2004). Consequently, molluscs are increasingly being analyzed using this technique, although so far primarily in studies focusing on selected aspects of shell and spicule morphology (Ruthensteiner et al. 2010, Alba-Tercedor and Sánchez-Tocino 2011, Connors et al. 2012, Lukeneder 2012, Sutton and Sigwart 2012, Ziegler and Menze 2013, Liew et al. 2014, Jochum et al. 2014, Noshita 2014, Hoffmann et al. 2015, Jochum et al. 2015, Vinther 2015, Xavier et al. 2015, Becker et al. 2016, Liew and Schilthuizen 2016, Cotton et al. 2017, Jochum et al. 2017, Sumner-Rooney and Sigwart 2017). But, µCT has also been used in experimental malacological research, for example in studies on burrow systems of wood-boring bivalves (Amon et al. 2015) or of a sediment-boring gastropod (Hale et al. 2015) as well as on the effects of ocean acidification on calcified structures in selected conchiferans (Keklikoglou et al. 2015, Chatzinikolaou et al. 2017, Howes et al. 2017). In addition, this X-ray-based imaging technique was employed to study functional and biomechanical aspects of molluscs, including the microstructure of the byssus (Frølich et al. 2015), eye anatomy in a chiton (Li et al. 2015) as well as in a bivalve (Palmer et al. 2017), and radular tooth morphology in a chiton (Ukmar-Godec et al. 2015). Recently, µCT systems of the latest generation are being used to study early ontogenetic stages of molluscs, for example in attempts to understand shell growth or the formation of suture patterns (Inoue and Kondo 2016, Lemanis et al. 2016, Noshita et al. 2016).

However, the relatively weak performance of X-ray tubes even in modern µCT systems means that scans at voxel resolutions in the nanometer scale will take significantly longer than scans with similar parameters conducted using SRµCT beamlines. For specific studies it might therefore be of interest to obtain scan time at one of these facilities. Compared to µCT scanners, SRµCT beamlines offer improved physical properties and are usually equipped with robotic sample holders that enable high-throughput scanning. However, a notable caveat of such analyses is the high potential for irreversible damage to the specimen's DNA, a consequence of the high X-ray doses emitted on SRµCT systems (Immel et al. 2016). Significantly less damage to DNA can be observed when conventional µCT or CT systems are used (Paredes et al. 2012, Hall et al. 2015, Immel et al. 2016, Wanek and Rühli 2016). In addition, due to technical limitations SRµCT is commonly performed using a FOV of only a few millimeters width and height (Zanette et al. 2014), which obviously complicates the analysis of larger specimens. One example for SRµCT -based research on molluscs is provided by a study focusing on embryonic shell formation in a gastropod (Marxen et al. 2008).

CT

CT was here performed using a system originally designed for human diagnostics. The scan of an adult specimen of the relatively large cephalopod Dosidicus gigas provides only a limited degree of soft tissue contrast (Fig. 4) if compared with results on Sepia omani derived from clinical MRI (Fig. 10A, B). However, the specific CT scanner settings used here still permit differentiation of selected internal soft and hard parts like the digestive gland (Fig. 4B) or ink sac and gladius (Fig. 4D), in addition to information about the exterior shape (Fig. 4C). Therefore, despite its limited application in malacological research so far, CT must be seen as an imaging modality that may provide a significant amount of valuable morphological data on larger molluscs, in particular when scanning systems of the latest generation are employed. Prominent examples for such modern CT scanners are interventional angiography systems used in human diagnostics, primarily during surgery. These scanners have a smaller FOV (e.g., 30 × 30 cm) than conventional CT systems, but can acquire 3D datasets at a relatively high IVR (e.g., 150-500 µm) in less than a minute.

Contrast-enhanced µCT

The µCT and CT scans presented here show that the attenuation contrast typically employed in X-ray-based imaging yields data primarily restricted to hard parts, even in the presence of large amounts of soft tissues (Fig. 1, Fig. 3A-D). Nevertheless, soft part contrast can be obtained using such techniques, but the soft tissues of the specimen under study need to be enriched with an X-ray-dense contrast agent (Metscher 2009, Pauwels et al. 2013). In the present study, PTA was chosen for contrast-enhanced µCT, because this chemical offers excellent contrast properties. However, PTA exhibits only a relatively slow diffusion presumably due to the large size of the compound. PTA was previously estimated to penetrate earth worm tissues at about 1 mm per week, but this parameter may depend on various chemical and physical properties of the specific tissues involved (Fernández et al. 2014). The results obtained here illustrate that PTA is a suitable contrast agent for molluscan specimens and can - given sufficient diffusion time and initial concentration - penetrate even thick tissues, as shown here by the successful staining of the foot of a chiton (Fig. 5A-C) and the mantle of a cephalopod (Fig. 5D-G). In contrast, smaller specimens with less dense tissues such as the cephalopod embryo analyzed here (Fig. 6) will stain rapidly. Prominent alternatives to PTA are the various compounds containing iodine (Metscher 2009, Pauwels et al. 2013, Gignac et al. 2016), but care needs to be taken when using these chemicals as they have been shown to lead to significant specimen shrinkage as well as overstaining under certain conditions (Vickerton et al. 2013, Buytaert et al. 2014, Gignac and Kley 2014, Li et al. 2015, Sombke et al. 2015, Li et al. 2016, Hedrick et al. 2018).

With regard to their usefulness in contrast-enhanced micro;CT, common to all staining agents mentioned above is that they do not display any form of specific staining properties (Metscher 2009, Pauwels et al. 2013, Gignac et al. 2016). However, although not a specific stain per se, certain contrast preferences of PTA have been documented in the past during studies aimed at using this agent in ultrastructural analyses (Quintarelli et al. 1971, 1973). However, the specific staining properties of PTA pale in comparison to established histological stains such as hematoxylin and eosin, Azan, or Mallory's trichrome. This important fact must be kept in mind when analyzing data derived from contrast-enhanced µCT experiments, because the resulting grey-scale images require a profound knowledge of molluscan anatomy to avoid misinterpretations. One way to facilitate the identification of different structures of complex µCT datasets is to apply dual-energy CT methods (Handschuh et al. 2017).

Studies that previously have analyzed molluscan taxa using contrast-enhanced µCT comprise experiments on the general suitability of all higher molluscan taxa for this approach using different contrast agents (Golding and Jones 2007, Faulwetter et al. 2012, Cándas et al. 2015, Faulwetter et al. 2015, Cándas et al. 2016). In addition, more specific studies have focused on odontophoral cartilage morphology in selected gastropods (Golding et al. 2009), changes in ganglionic system anatomy during cephalopod ontogeny (Kerbl et al. 2013), the internal anatomy of a larval cephalopod (Xavier et al. 2015), the reproductive system of a nudibranch (Cándas et al. 2017), the anatomy of Hoyle's organ in a cephalopod (Cyran et al. 2018), or the visualization of internal and external organs of selected gastropod species (Pedrouzo et al. 2018). Furthermore, contrast-enhanced µCT was performed in a methodological study on correlative imaging using a bivalve (Handschuh et al. 2013), to describe several new nudibranch species (Moles et al. 2016, Moles et al. 2017), and to illustrate the internal anatomy of numerous bivalve species (Machado et al. 2018).

As outlined above, contrast-enhanced µCT involves penetration of the specimen by an X-ray-dense contrast agent, thus altering the object under study. While such an approach may be applicable for certain samples, it constitutes an interference with the specimen's integrity. This will in principle render this protocol impractical for use on rare or otherwise valuable specimens, e.g., type material. However, initial results from a limited number of studies suggest that stain removal could be possible under certain conditions (Schmidbaur et al. 2015, Gignac et al. 2016). An entirely non-invasive alternative to contrast-enhanced µCT is phase contrast imaging. This modality involves technical adjustments to attenuation contrast-based SRµCT beamlines (Zehbe et al. 2010, Saccomano et al. 2018), but lately also to lab-based µCT equipment (Zanette et al. 2014). In molluscs, this approach has so far only been used to analyze pearls (Krzemnicki et al. 2017), but is bound to develop into a promising technique for malacological research in the near future.

Preclinical MRI

As shown in the previous µCT scans, excellent soft part information can be obtained using attenuation contrast-based techniques in combination with staining agents, but this approach may not be applicable in case of all molluscan specimens. As an alternative to this non-destructive, but invasive approach to visualizing molluscan structural detail in 3D, various MRI techniques can be employed. In contrast to µCT or CT, information on molluscan soft parts is usually obtained without the use of contrast agents, thus rendering this approach particularly attractive for the study of type material. In order to achieve high voxel resolutions, preclinical MRI systems are conventionally employed. This approach uses equipment designed for experimental animals such as mice, rats, or smaller primate species (Ziegler et al. 2011, Zanette et al. 2014). However, these instruments have a limited FOV and are therefore only suitable for specimens on the millimeter and centimeter scale. In addition, due to their lower overall distribution, preclinical MRI systems are usually less accessible than clinical MRI scanners (Ziegler et al. 2008). But, in contrast to the latter, there are no restrictions regarding protocol choice on preclinical MRI scanners (see below), although overheating of the gradient system may sometimes lead to an interruption of longer 3D scans (Ziegler et al. 2011). In addition, MRI in general shows strong artifacts when zoological specimens are analyzed that contain para- or ferromagnetic substances (Ziegler and Mueller 2011). Prominent examples include ingesta contained within the digestive tract (e.g., in sediment feeders) or the strongly mineralized radula of chitons (Ziegler et al. 2011). Despite these caveats, the scans conducted in the present study show that preclinical MRI must be considered as a powerful tool to visualize molluscan internal and external structures at voxel resolutions in the single- to double-digit micrometer scale obtained from uncontrasted millimeter- to centimeter-sized specimens.

Several studies on molluscan specimens have so far been conducted using preclinical MRI, including analyses of the single-cell and sub-cellular composition of neurons from derived taxa (Hsu et al. 1996, Grant et al. 2000, Gozansky et al. 2003, Lee et al. 2015), physiological capacities of cephalopods and bivalves measured in vivo (Bock et al. 2004, Melzner et al. 2006, Bock et al. 2007, Melzner et al. 2007, Lannig et al. 2008, Gutowska et al. 2010, Lee et al. 2010, Seo et al. 2014, Sukhotin et al. 2017), reproductive organ maturation and sex determination in oysters (Davenel et al. 2006, Davenel et al. 2010, Smith and Reddy 2012), or visual system anatomy in cephalopods (Jiang et al. 2013, Chung and Marshall 2014, Chung and Marshall 2017, Liu et al. 2017, Liu et al. 2018). In addition, preclinical MRI was applied to study the anatomy of a cobalt-labeled cephalopod brain (Quast et al. 2001), to conduct analyses of pathological alterations in belemnites (Mietchen et al. 2005, Mietchen et al. 2008) and to study the internal anatomy of a gastropod (Ziegler and Mueller 2011). In addition, the brain anatomy of a cephalopod (Solé et al. 2013), the morphology and adhesion mechanism of the octopus sucker (Tramacere et al. 2013), and the first 3D reconstruction of internal organs in a dumbo octopod hatchling (Shea et al. 2018) were performed using preclinical MRI.

With regard to specimen preparation for MRI, it is important to note that MRI protocols in principle permit scanning specimens fully immersed in ethanol (Sigl et al. 2013). This may be advantageous for various reasons, for example if the higher surface tension of water leads to air bubbles being trapped inside a specimen, which usually results in the presence of strong susceptibility artifacts (Ziegler et al. 2011) or when museum specimens have to be analyzed as non-invasively as feasible. However, the possibility to scan a specimen in ethanol depends on various parameters, in particular magnetic field strength and ethanol concentration. In general, a mix of liquids is problematic for MRI as such a solution may lead to chemical shift artifacts, e.g., between water and ethanol. For these reasons, all specimens analyzed here using preclinical MRI systems were either immersed in distilled water prior to scanning, were kept in their formalin fixative, or were embedded in agarose in order to provide a homogenous medium surrounding the specimen.

Clinical MRI

In contrast to the previous technique, clinical MRI is using scanning systems specifically designed for human diagnostics. Due to their technical optimization for medical examinations of living human subjects, clinical MRI systems usually employ fast scanning sequences, often inhibiting longer protocols with large volume coverage and high voxel resolutions, because such sequences might impact the patient due to rapid gradient switching or the applied radiofrequency power. However, the settings on most clinical MRI systems can be changed by trained personnel to accommodate zoological specimens intended for longer 3D scans. But, as illustrated here (Fig. 10A, B), 2D scanning sequences with resolutions along the Z axis that closely approach those in the X and Y axes (e.g., 500 × 500 × 800 µm) can also be used to obtain valuable structural information from molluscan specimens, whether in vivo or ex vivo. Clinical MRI studies that have analyzed molluscan taxa in the past include experiments on various morphological and anatomical features of bivalves (Pouvreau et al. 2006, Holliman et al. 2008, von Brand et al. 2009), soft part morphology in a gastropod (Cooper 2011) as well as selected cephalopods (Ziegler and Menze 2013, Xavier et al. 2015), the kinematics of buccal mass movement in a gastropod (Neustadter et al. 2002, Novakovic et al. 2006), and the auditory capacity of a cephalopod (Mooney et al. 2010).